On the regio- and stereoselective synthesis of aminocyclitols from cyclitol epoxides: The effect of Li as a chelating agent

Article abstract of DOI:10.1002/chem.200401270

Experimental and theoretical studies on the influence of Li ions on the regio- and the stereoselectivity of the reaction of cyclitol epoxides with nitrogen nucleophiles have been carried out. Model studies with NaN₃ as a nucleophile in the abse

Full text of DOI:10.1002/chem.200401270

FULL PAPER
On the Regio- and Stereoselective Synthesis of Aminocyclitols from Cyclitol
Epoxides: The Effect of Li as a Chelating Agent
Pedro Serrano,[a, b] Amadeu Llebaria,[b] Jordi Vµzquez,[c] Joan de Pablo,[c, d]
Josep M. Anglada,*[e] and Antonio Delgado*[a, b]
Abstract: Experimental and theoretical
studies on the influence of Li ions on
the regio- and the stereoselectivity of
the reaction of cyclitol epoxides with
nitrogen nucleophiles have been car-
ried out. Model studies with NaN3 as a
nucleophile in the absence of Li ions
predict a mixture of C1 and C2 re-
gioadducts. The inclusion of two Li
ions as a chelating agent favours the
operation of a low populated “all-
axial” conformation leading ultimately
to the C1 adducts. In all cases, the re-
sults can be rationalised by geometric
and energetic considerations of the cor-
responding transition states. Predic-
tions of the theoretical calculations are
in good agreement with the experimen-
tal results using primary and secondary
amines as nucleophiles, and thus con-
firm the validity of this study.
Keywords: ab initio calculations ·
amino alcohols
·
conformation
analysis · epoxides · regioselectivity
Introduction
Over the last years, aminocyclitols have gained interest as
pharmacological tools for the study of cellular processes
linked to the inositol phosphate cycle, as well as potential
glycosidase inhibitors.[1–3] As part of our ongoing research
directed towards the design and synthesis of new amino and
diaminocyclitols as modulators of sphingolipid metabolism,[4]
we became interested in the regio- and stereoselective syn-
thesis of aminocyclitol derivatives of types 3–5 from regiose-
lective epoxide opening of suitably protected cyclitol epox-
ides 1[5] and 2[6] (Scheme 1).
[a] Dr. P. Serrano, Dr. A. Delgado
Universidad de Barcelona, Facultad de Farmacia
Unidad de Química FarmacØutica (Unidad Asociada al CSIC)
Avda. Juan XXIII, s/n, 08028 Barcelona (Spain)
Fax : (+34)932-045-904
E-mail: adelgado@cid.csic.es
antonio.delgado@ub.edu
[b] Dr. P. Serrano, Dr. A. Llebaria, Dr. A. Delgado
Research Unit on Bioactive Molecules (RUBAM)
Departament de Química Orgànica Biològica
Institut d’Investigacions Químiques
i Ambientals de Barcelona (IIQAB-C.S.I.C)
Jordi Girona 18-26, 08034Barcelona (Spain)
[c] Dr. J. Vµzquez, Dr. J. de Pablo
Grup de Modelització Molecular
Centre Tecnològic de Manresa
Avda. Bases de Manresa 1, 08240 Manresa (Spain)
[d] Dr. J. de Pablo
Departament d’Enginyeria Química
Universitat Polit›cnia de Catalunya, Av. Diagonal 647
08028 Barcelona (Spain)
Scheme 1. Regio- and stereocontrolled opening of cyclitol epoxides.
[e] Dr. J. M. Anglada
Departament de Química Orgànica Biològica
Institut d’Investigacions Químiques
i Ambientals de Barcelona (IIQAB-C.S.I.C)
Jordi Girona 18-26, 08034Barcelona (Spain)
Fax : (+34)932-045-904
Examples of regiocontrolled epoxide opening of cyclitol
derivatives with nitrogen nucleophiles to give the corre-
sponding C1 or C2 adducts are scarce in the literature.[7] In
a previous work,[6] we described an unexpected regioselec-
tive chelation-controlled C1 azidolysis and aminolysis of ep-
oxide 1 in the presence of Yb(OTf)3. This method required
E-mail: anglada@iiqab.csic.es
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
Chem. Eur. J. 2005, 11, 4465 – 4472
DOI: 10.1002/chem.200401270
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4465

the presence of a vicinal free hydroxyl group for an efficient
metal coordination, which prevented their practical use on
fully protected cyclitol derivative 2. Along this line, we were
interested in the search of an alternative, more general pro-
tocol for this regioselective transformation.
Reactivity of cyclitol epoxides with nitrogen nucleophiles:
Based on these premises, epoxides 1 and 2 represent chal-
lenging models to confront the above mechanistic hypothe-
ses. Thus, intramolecular chelation would require the partici-
pation of an apparently highly energetic “all-axial” confor-
mation A1, in equilibrium with the equatorial conformations
E1a and E1b (Scheme 3). Both axial and equatorial confor-
mations would lead to different regioadducts on the basis of
the above-mentioned stereoelectronic effects. However,
treatment of epoxides 1 and 2 with NaN3 in the presence of
2n LiClO4 in CH3CN at 808C[15] cleanly afforded the corre-
sponding azidoalcohols 3a, and 4a in high yields from the
regio- and diastereoselective C1 opening of the starting ep-
oxides (see Table 1, entries 1 and 5).[16] The use of primary
or secondary amines as nucleophiles also afforded the corre-
sponding C1 regioadducts (Table 1, entries 2, 3, 4, 6, 7, and
8). The role of the CH3CN/LiClO4 system seems crucial for
the reaction outcome, as evidenced in entries 9 to 12, where
LiClO4 was suppressed and/or CH3CN was replaced with
THF.
Epoxide opening of substituted 1,2-epoxycyclohexanes is
well documented in the literature.[8] In general, a trans-dia-
xial stereoselectivity is found, as dictated by stereoelectronic
effects.[9–13] However, the observed regioselectivity depends
largely on conformational bias and can be modulated by a
judicious choice of reaction conditions. In this context,
model studies on the reactivity of cis- and trans-4-benzyl-
oxy-1,2-epoxycyclohexane with nitrogen nucleophiles have
shown the dramatic effect of some metals, particularly Li
salts, to promote a complete regiochemical epoxide open-
ing.[14] Thus, in the presence of LiClO4, the conformational
equilibrium in the cis isomer is shifted towards a reactive
conformation in which the benzyloxy group adopts an axial
disposition to enable intramolecular metal chelation with
the epoxide. On the other side, the conformational equilibri-
um in the trans isomer is also shifted to allow a lithium-as-
sisted regiocontrolled intramolecular nucleophile delivery
(Scheme 2).
The regioselective formation of C1 adducts 3 and 4 with
azide and amines can be interpreted as a result of an exclu-
sive trans-diaxial opening of epoxides 1 and 2 through the
“all-axial” Li-chelated confor-
mation A1 shown in Scheme 3.
Conformation A1 is expected
to be strongly stabilized by a
double intramolecular chelation
between Li and epoxide and/or
benzyloxy groups. This confor-
mation is consistent with the
observed regiochemical out-
come arising from the expected
stereocontrolled
opening.
trans-diaxial
Scheme 2. Effect of Li ions on epoxide ring opening in 4-benzyloxy-1,2-epoxycyclohexane.[14]
Scheme 3. Lithium promoted opening of cyclitol epoxides 1 (R=H) and 2 (R=Bn).
4466
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4465 – 4472

Cyclitol Epoxides
FULL PAPER
Table 1. Opening of epoxides 1 and 2 in the presence of 2n LiClO4.
The density functional theory
with the B3LYP functional[20]
has been used to compute the
different stationary points along
the reaction paths. The polar-
ized continuum model (PCM)
of Tomasi and co-workers[21,22]
has been employed to analyse
the solvent effects (acetonitrile,
e=36.64). All the energetic
values discussed below are also
based on results taking into ac-
count both the entropic correc-
Substrate
Nucleophile
Products
dr[a]
Yield[b] (%)
1
2
3
4
1
A
B
C
D
3a
3b
3c
3d
single
single
single
single
92
90
85
91
tions and the solvent effects.
Additionally, the bonding fea-
tures for the Li chelated struc-
tures have been analyzed with
5
6
7
8
2
A
B
C
D
4a
4b
4c
4d
(d)
(d)
4b/5b
4b/5b
single
single
single
single
–
–
2:3
2:3
91
94
92
95
–
–
87
83
the atoms-in-molecules theory
9
A[c]
A[e]
B[c]
B[e]
by Bader.[23] Herein we describe
10
11
12
only an schematic representa-
tion of the optimized structures
corresponding to reactants and
transition states, while the
[a] Determined by 1H NMR or HPLC. [b] Isolated yields. [c] No LiClO4
was added. [d] No reaction. [e] THF as solvent.
structures of the remaining in-
vestigated stationary points and
final products are available as
Attempts to detect the putative reactive conformation A1
Supporting Information.
1
by H or 13C NMR experiments were unsuccessful (see Sup-
For the sake of clarity, we have first modelled the process
without including the chelating agent. Thus, in the absence
of Li ions, only two stable conformations are found, namely
A2 and E2 with “all-axial” and “all-equatorial” substituents,
respectively (see Scheme 3 and Figure 1). Both conformers
adopt a pseudo-chair ring disposition and, according to our
calculations, they are energetically “quasi-degenerate” (E2
is computed to be more stable than A2 by only
0.8 kcalmolÀ1; DG value). In our study, we have considered
the nucleophilic attack by the azide ion at both C1 and C2
for each of the conformers and a free energy profile of the
corresponding reactions paths have been plotted in Figure 2.
In each case, the reaction starts with the formation of a van
der Waals pre-reactive complex between the epoxide and
the azide ion, prior to the transition state and the formation
of the corresponding products. Figure 2 also shows that the
most favourable reaction path for the nucleophilic attack
from A2 (all axial substituents) involves a trans-diaxial ep-
oxide opening process, with a computed free energy barrier
of 32.4kcalmol À1 (A2-TSC1) and, consequently, it would
give rise to the C1 adduct, while the attack at C2 requires a
higher free energy barrier of 36 kcalmolÀ1 (A2-TSC2). It is
also apparent from Figure 2 that, in the absence of Li ions,
the reaction is endoergic by 7.3 kcalmolÀ1 and that the pre-
reactive complex is unstable with respect to the reactants.
This can be interpreted as a result of a differential entropic
effect and it points out that the equilibrium is shifted to the
reactants side. Noteworthy our calculations predict the prod-
uct A2-P2, having a boat-like ring structure, to be about
4kcalmol À1 more stable than A2-P1, with a chair-like ring
porting Information). Thus, only minor changes on the
NMR pattern for the cycloaliphatic protons were observed
when a 0.02m solution of epoxide 2 in CD3CN was treated
with LiClO4 at concentrations ranging from 0 to 1.5m[17] and
temperatures from 25 to 608C. These results indicate that,
contrary to our own observations in model cis-4-benzyloxy-
1,2-epoxycyclohexane[14] (see also Supporting Information),
the conformational equilibrium of cyclitol epoxide 2 is not
altered in the presence of Li salts.[18] It seems thus reasona-
ble to assume that the major conformers in solution are the
equatorial ones (E1 and E2, Scheme 3) and that most of the
Li ion is coordinated with the solvent.[19] The major or exclu-
sive formation of C1 regioadducts can thus be explained
either by operation of the postulated “trans-diaxial” opening
from a low populated reactive conformation A1 or by an al-
ternative, formal “trans-diequatorial opening” from any of
the equatorial conformers, in clear conflict with the accepted
stereoelectronic control for these processes.[9–13]
Theoretical calculations: With the aim of gaining insight
into this intriguing mechanism, we have performed a com-
putational work on different reaction paths concerning the
possible formation of C1 and C2 adducts by a nucleophilic
attack on cyclitol epoxides 1 or 2, as shown in Scheme 3. In
order to have a computationally feasible system, a model
structure analogous to 1, in which the OBn groups have
been replaced with OCH3 groups has been considered.
Moreover, the nucleophilic attack on each of the epoxides
has been modelled by using the azide ion as a nucleophile.
Chem. Eur. J. 2005, 11, 4465 – 4472
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4467

A. Delgado, J. M. Anglada et al.
rier of 33.2 kcalmolÀ1 has been
found. The reaction path is
again endoergic by about
14kcalmol À1 and the pre-reac-
tive complexes are unstable
with respect to the reactants,
this pointing out that the equi-
librium between the reactants
and the pre-reactive complexes
is also shifted to the reactants
side.
The observed preference for
the nucleophilic attack on C1
or C2 depending on the axial
(A2) or equatorial (E2) disposi-
tion of the substituents can be
rationalized by considering the
geometries of the correspond-
Figure 1. Structure of the reactants. Bond lengths are in angstroms.
ing transition states (Figure 2).
In both cases, the transition
structure, in agreement with the large electronic repulsion
of the all-axial substituents in A2-P1. On the other side, the
free energy profile also displayed in Figure 2, shows that the
most favourable reaction path for the nucleophilic attack on
E2 (all equatorial substituents) would occur through E2-
TSC2, (with a computed free energy barrier of 28.2 kcal
molÀ1) to afford the C2 adduct (EP2), while for the corre-
sponding attack on C1 (E2-TSC1) a higher free energy bar-
states of lower activation barriers (A2-TSC1 and E2-TSC2,
respectively) present a chair-like structure, while those of
higher activation barrier present a more strained boat-like
structure (A2-TSC2 and E2-TSC1). In addition, the results
displayed in Figure 2 show that the chair-like transition
states are between 4–5 kcalmolÀ1 more stable than the cor-
responding boat-like ones, and this energy compares well
with the relative stability of a cyclohexane chair conforma-
Figure 2. Schematic free energy diagram for the nucleophilic attack of azide to A2 and E2. Each stationary point along the reaction path is labelled by
the letter of the reactant followed by a letter designing the van der Waals complex (C), the transition state (TS) or the product (P). In addition, a suffix
1 or 2 denotes that the nucleophilic attack takes place at C1 or C2, respectively. The structures of the transition states are also included and the bond
lengths are in angstroms.
4468
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4465 – 4472

Cyclitol Epoxides
FULL PAPER
tion with respect to the boat one (around 5.5 kcalmolÀ1).[24]
Thus, our theoretical calculations predict that the nucleo-
philic attack on the “all-axial” epoxide A2 takes place at C1
while attack on the “all-equatorial” epoxide E2 occurs at
C2, as a result of the more favourable ring conformation for
each transition state. Moreover, the small energetic differ-
ence between A2 and E2 predict the low regioselectivity
found for this process in the absence of Li ions. Our experi-
mental results are in agreement with these predictions
(Table 1, entry 11).
The effect of lithium as a chelating agent has been investi-
gated by including two Li ions.[25] The number of Li ions in-
cluded in the calculation has been done by taking into ac-
count the geometric disposition of the “all-axial” reactant
conformer, which allows one of the Li ions to simultaneous-
ly chelate to three different oxygen atoms from one side of
the molecule (the epoxide, the OH group and one of the
OMe groups, respectively), the second Li atom being able to
chelate to two different oxygen atoms (two OMe groups)
from the opposite side, this giving rise to the A1 chelated
conformer shown in Figure 1. The possibility of Li atoms to
simultaneously chelate to two or three oxygen atoms is sup-
ported by recent literature reports.[26–31] Regarding the “all-
equatorial” conformation, there are two possibilities for Li
chelation leading to E1a and E1b conformers, respectively
(see Scheme 3 and Figure 1). In each case, each Li atom
binds to two oxygen atoms, namely the epoxide and the OH
group from one face and the two OMe groups from the op-
posite face of the molecule. From an energetic standpoint,
the ability of one Li ion to simultaneously chelate to three
oxygen atoms in the “all-axial” conformation results in a sig-
nificant extra stabilization of A1 with respect to the two
“all-equatorial” chelated conformers E1a and E1b (DG=
13.64and 10.09 kcalmol À1, respectively), which shifts the
equilibrium to the “all axial” A1 conformer (Figure 3).
For each elementary process, the reaction path begins
with the formation of a van der Waals complex between the
corresponding Li-chelated conformer and the azide anion,
prior to the evolution to the corresponding transition states.
The effect of the Li ions is also remarkable, compared with
the reactions described above in the absence of Li, and pro-
duces a strong energetic stabilization of all stationary points
along the reaction path (compare the energetic profiles in
Figure 2 with those displayed in Figure 3). For the nucleo-
philic attack on A1 (“all-axial” substituents), our calcula-
tions show that the pre-reactive complexes (A1-C1 and A1-
C2) are 22.1 and 22.7 kcalmolÀ1 more stable than the reac-
tants and, as expected, the lower activation barrier
(10.77 kcalmolÀ1 relative to the pre-reactive complex A1-
C1) corresponds to the C1 attack, in agreement with the
chair-like structure of the corresponding transition state
(A1-TSC1) (see Figure 3). On the other side, C2 attack has
a higher free energy barrier (18.34kcalmol À1 relative to the
pre-reactive complex A1-C2), as expected from the boat-
like structure of the corresponding transition state A1-TSC2
(see Figure 3). The reaction path displayed in Figure 3
Figure 3. Schematic free energy diagram for nucleophilic attack of azide ion to A1 and E1b. The stationary points are named as in Figure 2 and the struc-
tures of the transition states are also included. Bond lengths are in angstroms.
Chem. Eur. J. 2005, 11, 4465 – 4472
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4469

A. Delgado, J. M. Anglada et al.
shows that both transition states lay energetically below the
reactants and the reaction is exoergic, being the C1 adduct
(A1-P1) the most stable product.[32]
of the corresponding transition states. In each case, a chair-
like transition state is energetically favoured in comparison
with the more strained boat-like ones. These predictions are
in agreement with our experimental results.
Despite the large stability computed for A1, and for the
sake of completeness, we have also investigated the nucleo-
philic attach of the azide ion upon C1 and C2 for the most
stable “all-equatorial” chelated conformer E1b. Free energy
profiles of the corresponding reaction path is also plotted in
Figure 3. The nucleophilic attack also starts with the forma-
tion of the corresponding van der Waals pre-reactive com-
plexes (E1b-C1 and E1b-C2) between the Li-chelated epox-
ide and the azide ion. Our calculations predict those pre-re-
active complexes to be more stable than the corresponding
reactants (DG = 16.73 and 16.86 kcalmolÀ1, respectively).
The inclusion of two Li ions as a chelating agent has a
dramatic effect. From an energetic standpoint, a remarkable
stabilization of all stationary structures along the reaction
paths has been observed when compared with the process in
the absence of Li ions. Moreover, the “all-axial” conformer
A1 has a greater ability to chelate Li ions than the “all-
equatorial” conformers E1a and E1b, which results in a
larger energetic stability of the former and a shift of the
conformational equilibrium towards the A1 side. In addi-
tion, nucleophilic attack upon A1 leads to the C1 adduct, as
expected from the chair-like structure of the corresponding
transition state. However, this reactive conformation is low
Unexpectedly, the lower activation barrier (DG
=
20.46 kcalmolÀ1 relative to the pre-reactive complex) corre-
sponds to the C1 attack, with a boat-like transition state
structure (E1b-TSC1). On the other hand, C2 attack shows
a higher free energy barrier (DG=24.58 kcalmolÀ1, relative
to the corresponding pre-reactive complex) despite the fact
that the corresponding transition state (E1b-TSC2) has a
chair-like structure. In addition, the reaction is computed to
be exoergic by about 17 kcalmolÀ1. The unexpected prefer-
ence for the C1 attack upon E1b, in comparison with the
preference for C2 attack observed for the also “all-equatori-
al”, although not Li-chelated, E2 (see Figure 2) can be ex-
plained by the effect of Li ions. Thus, as shown in Figure 3,
the nucleophilic attack at C2 takes place through a transi-
tion state (E1-TSC2) where both Li atoms chelate to two
different atoms: the OH group and the epoxide from one
face and one of the OMe groups and the azide terminal ni-
trogen atom from the opposite face. On the contrary, in
transition state E1-TSC1 (leading to the nucleophilic attack
at C1), one of the Li atoms is three-fold chelated with two
of the OMe substituents and with the azide terminal nitro-
gen atom. This gives rise to an extra stabilization that com-
pensates for the most favourable chair-like E1-TSC2 confor-
mation and accounts for the unexpected preference for C1
attack. These results let us conclude that, from a mechanistic
standpoint, the nucleophilic attack at C1 is the preferred
pathway for both “all-axial” and “all-equatorial” conformers
in the presence of Li ions, as experimentally observed.
1
populated at RT and not observable by H NMR spectrosco-
py due to the higher tendency of Li ions to coordinate with
acetonitrile, the solvent system. Our calculations also predict
that, unexpectedly, nucleophilic attack on E1b would also
lead to the C1 adduct, despite the apparently more strained
boat-like structure of the corresponding transition state.
This can be interpreted as a result of its higher ability for Li
chelation in comparison with the alternative chair-like tran-
sition state, which counterbalances the above effect. In both
scenarios, the theoretical calculations are in good agreement
with the experimental results, thus confirming the validity of
this study.
Experimental Section
For general experimental procedures, see Supporting Information. Epox-
1
ides 1[5] and 2[6] were prepared according to described procedures. H and
13C NMR spectra have been recorded in CDCl3 at 300 and 75.5 MHz, re-
spectively, unless otherwise stated.
Technical details of the calculations: All stationary points in the potential
energy surfaces described in this work have been fully optimized with the
hybrid density functional B3LYP method[20] by using the 6-31G(d,p) basis
set.[33] At this level of theory we have also calculated the harmonic vibra-
tional frequencies to verify the nature of the corresponding stationary
point (minima or transition state), to provide the zero point vibrational
energy (ZPE) and the thermodynamic contributions to the enthalpy and
free energy. Moreover, to ensure that the transition states connect the de-
sired reactant complexes and products, we have performed intrinsic reac-
tion coordinate calculations (IRC), plus geometry optimization at this
level of theory. The effect of the solvent has been taken into account by
performing single point calculations at the optimized geometries by using
the PCM model by Tomasi et al.[21,22] In this case, a dielectric constant of
36.64corresponding to acetonitrile has been used. These calculations
have been performed using the Gaussian 98 program package.[34]
Conclusion
We have carried out an experimental and a theoretical study
of the influence of Li ions on the reactivity of cyclitol epox-
ides with nitrogen nucleophiles aiming at rationalizing the
regio- and stereoselective formation of the corresponding
aminocyclitol derivatives.
The computational work on model studies with azide ion
as nucleophile in the absence of Li atoms predicts that the
nucleophilic attack upon the “all-axial” conformer A2 will
give rise to the C1 adduct while attack upon the “all-equato-
rial” conformer E2 leads to the C2 adduct. These results
may be rationalized by considering the geometric features
In several stationary points of interest, we have also analysed the bond-
ing features of the chelate by using the atoms in molecules (AIM) theory
by Bader.[23] To do this, we have employed the AIMPAC program pack-
age[35] to obtain the topological properties of the B3LYP/6-31G(d,p)
wave function. The Molden program[36] was also used to visualize the
geometrical and electronic features of the different stationary points. The
Cartesian coordinates and the absolute energies of all stationary points
are also available as Supporting information.
General synthetic methods and compound characterization—Reaction of
epoxides with nucleophiles in the presence of LiClO4: A solution of the
4470
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4465 – 4472

Cyclitol Epoxides
FULL PAPER
starting epoxide 1 and 2 (0.5 mmol) in CH3CN (9 mL) was added drop-
wise under argon over LiClO4 (80 mg, 0.75 mmol) at RT. A solution of
NaN3 (0.5 mmol) or the corresponding amine (see Table 1) in CH3CN
(1 mL) was next added and the reaction mixture was stirred at 808C
under argon. After 18 h, the reaction mixture was cooled to RT,
quenched with H2O (10 mL), extracted with CH2Cl2 (3ꢁ20 mL), and
dried over anhydrous Na2SO4. Filtration and evaporation afforded crude
aminoalcohols (See Table 1), which were purified as indicated below.
84.0, 85.0, 127.6, 127.7, 127.8, 127.9, 128.2, 128.3, 128.4, 128.5, 138.3,
138.4, 138.6; IR (film): n˜ =3412, 3017, 2984, 2972, 1465, 1436 cmÀ1
.
Acknowledgements
Financial support from Ministerio de Ciencia
y Tecnología, Spain
Compound 3a: Purified by flash chromatography (silica gel pretreated
(MCYT, BQU2002-03737 and BQU2002-04485-C02-01) and DURSI,
Generalitat de Catalunya (2001SGR00342, 2001SGR00085 and
2001SGR00048) are gratefully acknowledged. P.S. thanks Ministerio de
1
with Et3N) on elution with hexanes/EtOAc 1:1. H NMR (200 MHz): d =
2.78 (br, 1H), 3.46 (m, 5H), 3.67–3.99 (m, 6H), 7.20–7.35 (m, 15H); 13C
Educación, Cultura
y Deportes for a predoctoral fellowship (FPU-
NMR (50 MHz, C6D6): d = 53.5, 65.9, 73.0, 75.7, 82.4, 82.9, 127.8–128.7,
138.1; IR (film): n˜ = 3350 (br), 2109 cmÀ1
.
AP2000–0839). J.V. thanks “Torres Quevedo” Program (Ministerio de
Ciencia y Tecnología; MCYT, Spain) for financial support. The theoreti-
cal calculations carried out in this work have been performed at the
Centre de Supercomputació de Catalunya (CESCA), whose services are
gratefully acknowledged.
Compound 3b: Purified by flash chromatography (silica gel pretreated
1
with Et3N) on elution with hexanes/EtOAc 1:1. H NMR: d = 0.90 (t, J
= J’ = 7.5 Hz, 3H), 1.26–1.47 (m, 4H), 2.41 (t, 2H, J=9.4Hz), 2.69 (t,
2H, J=6.9 Hz), 3.40–3.59 (m, 4H), 4.81–4.94 (m, 6H), 7.25–7.38 (m,
15H); 13C NMR: d = 13.9, 20.3, 32.7, 44.5, 61.4, 71.3, 75.4, 75.7, 82.8,
84.3, 127.7, 127.8, 128.0, 128.4, 128.5, 138.5; IR (film): n˜ = 2920, 1499,
1451 cmÀ1; MS: m/z: 506 [M+H]+.
[1] V. H. Lillelund, H. H. Jensen, X. F. Liang, M. Bols, Chem. Rev. 2002,
102, 515–553.
[2] A. Berecibar, C. Grandjean, A. Siriwardena, Chem. Rev. 1999, 99,
779–844.
[3] B. V. L. Potter, D. Lampe, Angew. Chem. 1995, 107, 2085–2125;
Angew. Chem. Int. Ed. Engl. 1995, 34, 1933–1972.
[4] A. Delgado, P. Serrano, in XII Congreso Nacionalde la Sociedad Es-
paÇola de Química TerapØutica, Book of Abstracts, Sevilla, Spain,
2001, p. 61.
[5] C. Jaramillo, J. Chiara, M. Martín-Lomas, J. Org. Chem. 1994, 59,
3135–3141.
[6] P. Serrano, A. Llebaria, A. Delgado, J. Org. Chem. 2002, 67, 7165–
7167.
[7] For C2 regioselective opening with NaN3 of related O-protected cy-
clophellitol derivatives, see: a) K. Tatsuta, S. Miura, Tetrahedron
Lett. 1995, 36, 6721–6724; b) S. Ogawa, N. Chida, T. Suami, J. Org.
Chem. 1983, 48, 1203–1207; for reactions of cyclitol epoxides with
amines, see: c) R. A. Cadenas, M. Y. Grass, J. Mosettig, M. E. Gelpi,
Nucleosides Nucleotides 1990, 9, 21–34; d) S. Leicach, M. E. Gelpi,
R. A. Cadenas, Nucleosides Nucleotides 1994, 13, 2051–2058.
[8] M. Bartok, K. L. Lang, in Small Ring Heterocycles (Ed.: A. Hass-
ner), Wiley, New York, 1985.
Compound 3c: Purified by flash chromatography (silica gel pretreated
1
with Et3N) on elution with hexanes/EtOAc 1:1. H NMR: d = 1.09 (t, J
= J’ = 7.5 Hz, 6H), 2.57 (t, 1H, J = J’ = 9.2 Hz), 2.76 (m, 4H), 3.48
(m, 5H), 4.83–4.98 (m, 10H), 7.24–7.40 (m, 25H); 13C NMR: d = 15.1,
63.8, 70.3, 75.3, 75.5, 82.8, 84.5, 127.6, 127.7, 127.9, 128.3, 138.4, 138.5; IR
(film): n˜ = 3418, 1605, 1492 cmÀ1; MS: m/z: 506 [M+H]+.
Compound 3d: Purified by flash chromatography (silica gel pretreated
with Et3N) on elution with hexanes/EtOAc 1:1. 1H NMR: d = 2.40 (m,
1H), 2.77 (t, J=7.2 Hz, 3H), 2.99 (t, J=7.8 Hz, 2H), 3.43 (m, 4H), 4.7–
5.0 (m, 6H), 7.2–7.4(m, 20H); 13C NMR: d = 36.9, 46.1, 61.4, 71.5, 75.3,
75.7, 82.8, 84.2, 126.1, 127.7, 127.8, 127.9, 128.2, 128.4, 128.6, 138.3, 138.5,
139.7; IR (film): n˜ = 3575, 3421, 3087, 2923, 2856, 1953, 1879, 1741,
14396, 1454 cmÀ1
.
Compound 4a: Purified by flash chromatography (silica gel pretreated
with Et3N) on elution with hexanes/EtOAc 2:1. [a]D = À6.2 (c = 0.95,
CHCl3); 1H NMR: d = 3.40–3.44 (m, 4H), 3.50–3.65 (m, 2H), 4.75–4.96
(m, 8H), 7.26–7.35 (m, 20H); 13C NMR: d = 45.5, 66.4, 72.7, 75.6, 75.8,
75.9, 81.1, 82.4, 82.6, 83.5, 127.6, 127.7, 127.8, 127.9, 128.1, 128.4, 128.6,
137.9, 138.0, 138.1; IR (film): n˜ = 3448, 3107, 1456, 1359, 1200 cmÀ1
.
Compound 4b: Purified by flash chromatography (silica gel pretreated
with Et3N) on elution with hexanes/EtOAc 2:1. [a]D = +7.4( c = 1.5,
CHCl3); 1H NMR: d = 0.89 (t, J = J’=8.5 Hz, 3H), 1.21–1.42 (m, 4H),
2.4–2.8 (m, 3H), 3.20–3.45 (m, 5H), 4.80–5.02 (m, 8H), 7.26–7.34 (m,
20H); 13C NMR: d = 13.9, 20.3, 32.8, 45.9, 61.6, 71.7, 75.1, 75.7, 75.8,
80.3, 83.0, 84.0, 85.0, 85.5, 125.6–128.5, 138.1–138.3; IR (film): n˜ = 3418,
3062, 3028, 1669, 1492 cmÀ1; MS: m/z: 596 [M+H]+.
[9] D. K. Murphy, R. L. Alumbaugh, B. Rickbom, J. Am. Chem. Soc.
1969, 91, 2649–2653.
[10] a) E. L. Eliel, N. L. Allinger, S. J. Angyal, G. A. Morrison, in Con-
formationalAnaylsis , Wiley, New York, 1965, p. 352; b) P. Deslong-
champs, Stereoelectronic effects in Organic Chemistry, Pergamon
Press, Oxford, 1983; c) J. Valls, E. Toromanoff, Bull. Soc. Chim. Fr.
1961, 758–764.
[11] F. H. Newth, Q. Rev. Chem. Soc. 1959, 13, 30–47.
[12] S. J. Angyal, Chem. Ind. (London) 1954, 1230–1231.
[13] A. Furst, P. A. Plattner, Helv. Chim. Acta 1949, 32, 275–283.
[14] F. Calvani, P. Crotti, C. Gardelli, M. Pineschi, Tetrahedron 1994, 50,
12999–13022, and references therein.
Compound 4c: Purified by flash chromatography (silica gel pretreated
with Et3N) on elution with hexanes/EtOAc 2:1. [a]D = +21.5 (c = 1.75,
CHCl3); 1H NMR: d = 1.11 (t, J=6 Hz, 6H), 2.55 (m, 1H), 2.75 (q, J=
6 Hz, 4H), 3.18 (m, 1H), 3.51 (m, 2H), 3.63 (m, 2H), 4.6–5.0 (m, 8H),
7.2–7.4(m, 20H); 13C NMR: d = 15.7, 63.9, 69.7, 74.2, 74.9, 75.8, 75.89,
79.0, 82.8, 84.3, 85.7; IR (film): n˜ = 3421, 3058, 3028, 2906, 2897, 1659,
1493, 1456 cmÀ1; EI-HRMS: m/z: calcd for C38H45NO5: 595.3297; found:
595.3321.
[15] Experiments at RT failed to afford the opening adducts.
[16] In all cases, the regio- and the stereochemistry of the major adducts
was confirmed from selected H–H coupling constants from azidoal-
cohols 3a and 4a. Moreover, in the case of 3a, symmetry considera-
tions imposed by the meso nature of the azidoalcohol arising from
trans-diaxial C1opening of epoxide 1 were conclusive.
[17] Higher concentrations of LiClO4 led to precipitation.
[18] A conformational shift should alter the 1H NMR pattern of the cy-
clohexane moiety, since dihedral angles should be significantly dif-
ferent for both limit conformations (A1 and E1). Thus, high J values
would be indicative of any of the “all-equatorial” conformations,
whereas small J values would be expected for the “all-axial” one.
[19] E. M. Cabaleiro-Lago, M. A. Ríos, Chem. Phys. 2000, 254, 11–23.
[20] A. D. Becke, J. Chem. Phys. 1993, 98, 5648–5652.
[21] J. Tomasi, B. Mennucci, E. Canc›s, J. Mol. Struct. 1999, 464, 211–
226.
Compound 4d. Purified by flash chromatography (silica gel pretreated
with Et3N) on elution with hexanes/EtOAc 2:1. 1H NMR: d = 3.4–4.0
(m, 8H), 4.7–5.0 (m, 8H), 7.2–7.4 (m, 25H); 13C NMR (50 MHz, CDCl3):
d = 46.9, 51.6, 75.1, 75.2, 75.8, 75.9, 81.2, 83.1, 83.9, 84.9, 127.1, 127.4,
127.5, 127.7, 128.1, 128.2, 129.3, 129.4, 129.5, 138.2, 138.3, 138.4, 138.6,
140.2; IR (film): n˜ = 3416, 3923, 2912, 1426, 1455, 1357, 1245 cmÀ1; [a]D
= À4.2 (c = 2.85, CHCl3).
Compound 5b: Purified by flash chromatography (silica gel pretreated
with Et3N) on elution with hexanes/EtOAc 2:1. 1H NMR: d = 0.83 (t,
J=7.5 Hz, 3H), 1.25–1.40 (m, 4H), 2.49 (t, J=10.2 Hz, 1H), 2.56 (m,
1H), 268 (m, 1H), 3.34–3.63 (m, 5H), 4.6–5.0 (m, 8H), 7.2–7.4 (m, 20H);
13C NMR: d = 13.9, 20.3, 32.9, 45.9, 61.5, 71.7, 75.2, 75.7, 75.8, 80.3, 83.0,
Chem. Eur. J. 2005, 11, 4465 – 4472
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4471

A. Delgado, J. M. Anglada et al.
[22] B. Mennucci, J. Tomasi, J. Chem. Phys. 1997, 106, 5151–5158.
[23] R. F. W. Bader, Atoms in Molecules. A Quantum Theory, Clarendon
Press, Oxford, UK, 1990.
[31] L. A. Paquette, J. Tae, E. R. Hickey, R. D. Rogers, Angew. Chem.
1999, 111, 1502–1505; Angew. Chem. Int. Ed. Angew. Chem. Int.
Ed. Engl. 1999, 38, 1409–1411.
[24] J. Clayden, N. Greeves, S. Warren, P. Wothers, Organic Chemistry,
1st ed., Oxford University Press, Oxford, 2001.
[32] Experimental conditions require heating at 808C to revert the pre-
ferred coordination of Li ion with acetonitrile. (see ref. [19]).
[33] P. C. Hariharan, J. A. Pople, Theor. Chim. Acta 1973, 28, 213–222.
[34] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.
Robb, J. R. Cheeseman, V. G. Zakrzewsk, J. J. A. Montgomery, R. E.
Stratmann, J. C. Buran, S. Dapprich, J. M. Millam, A. D. Daniels,
K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi,
R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Och-
terski, G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, N. Rega,
P. Salvador, J. J. Dannenberg, D. K. Malick, A. D. Rabuck, K. Ra-
ghavachari, J. B. Foresman, J. Cioslowski, V. Ortiz, A. G. Baboul,
B. B. Stefanov, A. L. G. Liu, P. Piskorz, I. Komaromi, R. G. L.
Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Na-
nayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, M. W. W. W.
Chen, J. L. Andres, C. Gonzalez, M. Head-Gordon, E. S. Replogle,
J. A. Pople, Gaussian, Inc., Pittsburgh PA, 2002.
[25] The possibility of considering only one Li atom as chelating agent in
the theoretical calculations was ruled out since the experimental
conditions required the use of a saturated LiClO4 solution. On the
other side, the above experimental requirements would be consis-
tent with the participation of three Li ions that could chelate the
“all-equatorial” conformer. This possibility was taken into account
and despite the corresponding transition states for the C1 and C2
attack have been found, we could not obtain the optimised geome-
tries for the corresponding pre-reactive complexes. However, form
an energetic standpoint, the transition state describing the C1 attack
also lie below in energy than that corresponding to the C2 attack, in
agreement with the experimental findings.
[26] A. Streitwieser, M. Husemann, Y.-J. Kim, J. Org. Chem. 2003, 68,
7937–7942.
[27] L. A. Paquette, C. S. Ra, J. C. Gallucci, H.-J. Kang, N. Ohmori, M. P.
Arrington, W. David, J. S. Brodbelt, J. Org. Chem. 2001, 66, 8629–
8639.
[35] R. F. W. Bader, http://www.chemistry.mcmaster.ca/aimpac, down-
loaded May 2002.
[36] G. Schaftenaar, CAOS/CAMM Center, The Netherlands, 2002.
[28] L. A. Paquette, J. Tae, J. Am. Chem. Soc. 2001, 123, 4974–4984.
[29] L. A. Paquette, J. Tae, J. C. Gallucci, Org. Lett. 2000, 2, 143–146.
[30] J. Tae, R. D. Rogers, L. A. Paquette, Org. Lett. 2000, 2, 139–142.
Received: December 10, 2004
Published online: May 13, 2005
4472
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 4465 – 4472